Frequent cksum failure in arch copy & recovery

[Arjun Dontamsetti]

Hi,

We have a setup which copies oracle arch files from one machine (different location) to another using automatic arch copy script and then there is a recovery script.

In arch copy script we are using ftp (tried rcp also for remote copy).

The problem is files byte size looks same but cksum is different so the recovery script fails which runs cksum of the file on both machines before it proceeds.

For this archcopy alone I have setup a separate network with 100 HD and arch copy moved to that network. Even then the problem persists.

This cksum failure between the files happens sometimes quite frequently after every 10 files.

Is there a problem with ftp or rcp in 11.0 ?.
Never experienced such issues in my earlier assignments.

Thanks in advance for your expert advises

Thanks,

Arjun

[Stefan Farrelly]

Ive seen ftp problems like this before and theyre almost always on 100Mbit networks. The problem usually being that the network device the HP 100Mbit port is plugged into is not hard set to 100 HD - its set to auto negotiation. As a result once the network gets a little busy all sorts of strange ftp timeouts/ failures/ bad copies take place, but when the network is nice and quiet it works fine. Try setting your network hub/switch port to 100HD (if it isnt already).

[A. Clay Stephenson]

No known problems like you describe with ftp or rcp. Since the file sizes are correct, I think we can eliminate the transfer of busy files. I would use netstat -s to look for bad statistics. My best guess is that either you have a speed/duplex mismatch between one of the hosts and the switch. I would hard set both ends of both connections. The other thing I would look at is more than one interface on the same subnet.